Detect one file in a folder and delete everything i that folder.

Entropy1024 · 12-17-2016, 05:53 AM

This is a bit convoluted and I will do my best to explain it. Hope it makes sense.

I have written a bash script that will backup my blog using wget. It creates a tar.gz archive then a series of parity files then finally an MD5. This is then stored in a unique folder with the date under a master folder called blogbackup.

Sometimes because of connectivity issues the backup will fail and I only get a partial backup. I know the total compressed archive should be approx 220meg. As the blog will only get lager I would like to delete any backups less than 220meg. This way I wont be archiving any failed backups.

I can easily search for tar.gz files less than 220meg in all subfolders under blogbackup, however it will leave the parity and MD5 files.

What I need to figure out is how to detect a folder with a tar.gz less then 220Meg and delete EVERYTHING in that folder.

Is there a way to achieve this?

Many thanks
Tim

hazel · 12-17-2016, 06:02 AM

You could try du. Run the output through awk, looking for "tar.gz" plus <220M in field $1.

Habitual · 12-17-2016, 07:46 AM

Quote:

Originally Posted by Entropy1024

I can easily search for tar.gz files less than 220meg in all subfolders under blogbackup, however it will leave the parity and MD5 files.

You don't sound very certain of much.
Have you considered testing the archive or some other logic not-based on "approx 220meg"?
I'd like to see some code.
Something is missing.

allend · 12-17-2016, 09:57 AM

Given

Code:

bash-4.4$ tree -s blogbackup/
blogbackup/
├── [       4096]  20161218A
│** └── [          0]  a.tar.gz
├── [       4096]  20161218B
│** └── [     252268]  b.tar.gz
└── [       4096]  20161218C
    └── [          0]  c.tar.gz

3 directories, 3 files

then in a bash shell

Code:

bash-4.4$ cd blogbackup/
bash-4.4$ for f in */*.tar.gz; do (( $(stat -c %s $f) < 225000 )) && echo ${f%/*}; done
20161218A
20161218C

So the code is doable, but I am with Habitual, there is likely to be a better way.

nodir · 12-18-2016, 11:06 AM

find comes to mind to search for sizes:

Code:

find bu_folder -size -220M  -name '*tar.gz'  -

combined with either dirname or parameter expansion to get the path to the directory in which said files are found:
with dirname (the less sexy solution :-) )

Code:

find bu_folder -size -220M  -name '*tar.gz'  -exec sh -c '
        for f; do 
            echo $(dirname "$f")
        done' echo_failed_bu {} \;

getting all files in that folder:

Code:

find bu_folder -size -220M  -name '*tar.gz'  -exec sh -c '
        for f; do 
          for i in $(dirname "$f")/*; do 
            echo "$i"
          done
        done' echo_failed_bu {} \;

In case you think that is a solution to your liking, look here to get the exact syntax you need for find: http://mywiki.wooledge.org/UsingFind
and here to use parameter expansion to get the path of the file you found: http://mywiki.wooledge.org/BashFAQ/0...meterExpansion
If the final ; doesn't work, you may want to try a + instead.

JeremyBoden · 12-18-2016, 11:38 AM

A different option would be to build your backup file with a suffix of .tmp
When it finishes, rename your file to drop the .tmp suffix.

Anything with a .tmp suffix is obviously a failed backup.

ondoho · 12-18-2016, 12:30 PM

Quote:

Originally Posted by Entropy1024

Sometimes because of connectivity issues the backup will fail and I only get a partial backup.

can't you run the backup directly on the server and its OS?
then connectivity wouldn't be an issue.